655 Stat 'Glitch' and Its Place in Tiering

[drampa's grandpa]

This thread is discussing the mechanic DaWoblefet posted about here. Check out that link for a more in-depth and technical explanation of what exactly is going on with this mechanic.

It was recently discovered that a Pokemon with 655 or more as its raw stat prior to modifiers has that stat roll back over to 0. Because of the way this interacts with the damage formula Pokemon with 0 Defense / SpD take almost no damage (2 before modifiers). This is clearly not balanced in any meaningful competitive way.
To clarify: only exactly 655 as the stat will take minimal damage. One more and you start taking massive damage due to having 1 Def/SpD.

While this won't affect standard tiers as the minimum stat to reach this threshold is 249, it does have an effect on several Other Metagames.

The affected tiers include...

Pure Hackmons - Eternatus-Eternamax, the Pokemon this was first discovered on, reaches the requisite defensive stats.

Tier Shift - Several Pokemon can in theory reach the requisite stat, but the only one currently around or that will likely ever drop far enough is Shuckle.

Mix and Mega - A number of Pokemon can reach the requisite defensive stats with either Slowbronite or Sablenite (or for Shuckle, various other stones). These include: Aggron, Avalugg, Cloyster, Regirock, Regice, Shuckle, Stakataka, Steelix. Not all of these exist at the current moment but most will in two days. This is the most popular and broadly affected of the metagames listed, besides perhaps Pure Hackmons.

Kartana can reach the requisite Attack stat. While I don't think it's been directly researched, so far it seems likely that this will be incredibly incredibly awful and do 2 damage per hit. The following mons could also in theory but are either banned, item-locked, non-existent at the moment, or some combination thereof, and as such are only included for completeness's sake.

Scalemons - Onix and Shuckle break the Defensive barrier, Feebas and Magikarp go over on Speed. I don't think there's anything else

Cross Evolution - Onix x Golisopod hits a high enough Defense stat.

350 Cup - Nothing right now, but Cranidos hits a high enough Attack stat should it be released.

Various OM Mashups - I'm not going to go into all of these right now, but different combinations of stat changing metagames will push various Pokemon to 655.

It's possible I missed something, but this isn't the main point. I simply wanted to show that a number of OMs are affected.

So the question arises: how do we tier this?

This has been a topic of discussion for several days in the Other Metas room. It seems to have boiled down to several different opinions:

1. Accept the mechanic and tier individual elements of the metagame around it
   1. Accept the mechanic an focus on banning individual abusers or methods of abuse of the mechanic
   2. Accept the mechanic and complex ban specific combinations that have the potential to abuse the mechanic (where applicable)
2. Accept the mechanic but clause it out
3. Completely ignore the mechanic
   1. Ignore the mechanic to the point of not coding it onto PS
   2. Ignore the mechanic by artificially removing it from the metagame (similar to 2)

My thoughts and analysis of each of these

3-1 is antithetical to Smogon tiering philosophy as it fails to accurately replicate cartridge mechanics, and should not be considered.

3-2 is somewhat of a departure from standard tiering procedures. However it has the benefit of being comparatively simple to understand, does not make it more difficult for new users to enter the metagame, and overall minimizes the impact of this discovery (which, IMO, is desirable). It does create a precedent beyond Sleep and Freeze Clause for modification of cartridge mechanics, and breaks the precedent against the same however.

2 has pretty much the same impact as 3-2. The main differences are that it is in effect a ban, not a modification of cartridge mechanics, and that it is slightly less easy to understand. However the tiering policy and competitive impact would likely be negligible as a whole, and it is unlikely that anyone will run across this issue without specifically attempting to abuse it. This is my preferred option.

1-2 is a complex ban, which isn't great. Overall it comes across as much more meta-changing, as it explicitly prevents a number of otherwise viable combinations, most notably on Shuckle in Mix and Mega. While it does avoid blanket bans, it is neither the simplest nor the most effective option.

1-1 This is the most obvious and straightforward solution, and what I would explicitly like to avoid having to do. It would make a number of Pokemon in a variety of tiers auto-banned, regardless of how healthy or unhealthy they otherwise are, and create a constant need for tiering action around it. "Methods of abuse" is meant to imply aspects of metagames such as the mega stones from Mix and Mega and the evolutions from Cross Evolution.

What I am asking for here is: what is the best option? Is there a way forward not encapsulated by what is laid out here? Please share your opinions, and let me know if anything I've said isn't clear enough (or if I was too long-winded :P). Everyone who has access to this forum is naturally welcome to provide their input, but I am especially looking for people who either have experience with the impacted tiers or people with experience tiering in general.

shoutouts Hogg, The Immortal, E4 Flint, and Ransei for giving me advice on this

 

[Plague von Karma]

I think Options 1-1 and 2 seem like the more likely solutions. The mechanic exists, and part of tiering policy is adapting to findings as they come. Gen 8 has been a ruling nightmare since release, this just seems to be another one. I'll explain why I prefer the options specified.

Option 1-1 feels like the better option over 1-2, as complex bans add rules, which is partly why they're disliked in the first place. Banning Pokemon adds to a list, which is easier to digest for the outsider. Generally, 1-2 looks like it would end up becoming a long list of complex bans, given what's been put forward so far. 1-1, on the other hand, falls more in line with traditional tiering policy. However, from what you're saying, this would end up being a "casualty of policy", thus the post in the first place...

In which case, though, why not take a similar route to Endless Battle Clause, where you have a singular rule which bans anything that could lead to an issue occurring? This is where Option 2 seems like it could be a stronger choice, banning Pokemon with defensive stats over 655 and requiring their stats to be decreased through levels, EVs and IVs. This could theoretically preserve accessibility in tandem with a proper PS warning, while producing the desired result of 1-2. In addition, Pokemon wouldn't need to be banned. This would also make it easier to apply to other OMs, since you won't have a list of complex bans on multiple rulesets that make no sense to the outsider. It doesn't look like a large departure from traditional tiering policy, either. It's still a complex ban by definition, but it seems to be a much easier route to go down than 1-1 or 1-2. As a bonus, it's naturally future-proofed, reducing the need for adaptation in the case new abusers of the mechanic slip through the cracks.

On the note of the Option 3s, I don't like either at all.

• Option 3-1, as you've said, is a no-go.
• Option 3-2 is a no-go simply on the basis that Pure Hackmons operates on having no rules at all, except for Endless Battle Clause, which has historically been used to prevent the PS servers from lagging. As such, this would cause a large inaccuracy to how it will be played on cartridge in the future. While you could say "3-1 applies to the other OMs" in this case, a lack of implementation would still hurt Pure Hackmons, which has had spotlight ladders and tours in the past. I just don't see this working.

Overall, a "Defense Overflow Clause" or something of that nature from Option 2 sounds like the best solution, being future-proofed and reducing the need for a long list of complex bans.
If I'm missing a point or something, well, feel free to tear this apart

 

[Merritt]

The least intrusive (but absolutely complex) ban of the actually broken piece of this glitch would be 'a Pokemon with exactly 596 in a defensive stat before the nature boost cannot have a positive nature' unless I've misunderstood how this works pretty badly. This is because 595 before nature does not overflow and 597 overflows to 1, not 0 (the problematic value).

I don't see any good reason to ban/ignore the overflow mechanic itself, it's a legitimate glitch in the same way GSC marowak sometimes catches new players off guard by doing almost 0 damage after SD.

Main thing is that this seems really annoying to program for mix and mega.

 

[drampa's grandpa]

I am also [sic] unaware of any situation where it is possible to have a stat above 654 but not possible to have a stat of exactly 654 by reducing EVs.

With this in mind, and the fact that basically everybody is in agreement that this interaction is blatantly broken and would need to be banned in any format it was possible in, I see no reason why we shouldn't simply have Showdown resolve any stat above 654 as exactly 654, and warn players of this cap before the match if they've built a team that otherwise exceeds it.

Posting on behalf of Rarycaris, as I think it's an interesting idea and certainly one I had not entertained before.

 

[DaWoblefet]

The least intrusive (but absolutely complex) ban of the actually broken piece of this glitch would be 'a Pokemon with exactly 596 in a defensive stat before the nature boost cannot have a positive nature' unless I've misunderstood how this works pretty badly. This is because 595 before nature does not overflow and 597 overflows to 1, not 0 (the problematic value).

I want to point out this is correct; a necessary condition of this overflow is having a boosting Nature. It should also in theory be possible to overflow with a hindering Nature as well based on the USUM source code, as that multiplication (*90 / 100) could get the numbers high enough. It's untestable because base stats cap at 255 on cartridge, but if you were able to reach 729 stat and then add the hindering Nature, that would also be sufficient to cause overflow (note: requires a base stat of at least 315). There's no placeholder *100 / 100 or anything for neutral natures so it can't be done that way.

Kartana can reach the requisite Attack stat. While I don't think it's been directly researched, so far it seems likely that this will be incredibly incredibly awful and do 2 damage per hit. The following mons could also in theory but are either banned, item-locked, non-existent at the moment, or some combination thereof, and as such are only included for completeness's sake.

Attack and Special Attack should also similarly overflow. It's impossible to test in Sword and Shield without ROM hacking, due to no Pokemon having high enough stats otherwise to reach the limit. However, we do know from other sorts of tests in USUM that 0 Attack also similarly causes base damage to become 2 before modifiers, and there's no special reason why Attack would be different from Defense behavior in SwSh.


As for the implementation of the mechanic itself: besides Pure Hackmons, I don't see any reason you can't justifiably mod the overflow out of OMs. They aren't following cartridge mechanics anyway, so I don't see why it'd be an issue to further deviate from cart to avoid these sort of unwanted scenarios.

 

[SadisticMystic]

The least intrusive (but absolutely complex) ban of the actually broken piece of this glitch would be 'a Pokemon with exactly 596 in a defensive stat before the nature boost cannot have a positive nature' unless I've misunderstood how this works pretty badly. This is because 595 before nature does not overflow and 597 overflows to 1, not 0 (the problematic value).

While 597 -> 1 (rather than 656) is not immediately busted, such a Pokemon that has a means of reducing its own Defense (like Close Combat) would get to count as 0 by going to -1, so the potential for "infinitely high defense" behavior is still there at slightly higher values.

 

[Zarel]

I don't understand the difference between 2 and 3-2. But that seems like the obvious choice to me.

 

[The Immortal]

Thanks for the feedback. An Overflow Stat Clause has been implemented, which caps stats at the value before they overflow to zero (i.e. 654 if running a positive nature, or 655 if running a negative nature).